Servo Health Monitoring Based on Feature Learning via Deep Neural Network

As the core actuator of an aircraft's flight control system, the servos' reliability directly affects the safety of the flight control system and the whole aircraft. The failure of the rudder will lead to the poor control effect of aircraft, affect its flight quality and safety, and even cause major flight accidents. In order to monitor the health status of servo and determine the fault and its degree accurately, this paper presents a feature learning based health monitoring method using a deep neural network. Firstly, we combine the wavelet packet decomposition and support vector machine to synthesize the sample segment label. And then, the sliding window is employed to enlarge the sample size, and the auto-encoder is utilized to reduce the data dimension. Moreover, the Softmax classifier is used for health monitoring. At last, the numerical simulations demonstrate the effectiveness of the proposed method.